Experimental Record for A Conservation Law for Commitment in Language Under Transformative Compression and Recursive Application (EXP-001 to EXP-007)

This record archives the complete experimental program accompanying A Conservation Law for Commitment in Language Under Transformative Compression and Recursive Application. Across EXP-001 to EXP-007, the series investigates whether commitment persists when language is subjected to recursive paraphrase, compression, gating, adversarial variation, mechanism isolation, self-application, and NP-negation edge-case testing. The goal of the program is not merely to show that outputs can remain superficially similar, but to examine whether identity-relevant commitment content survives transformation even when wording, syntax, or representational form changes. In that sense, the archive functions as the DOI-backed empirical companion to the law paper: a frozen evidence trail for the central claim that commitment persists through transformation, even when its form changes.

The experiments document a progressively refined empirical arc. Early runs establish the initial phase signal and show that recursive transformation does not simply dissolve commitment into noise. Later runs expand the corpus, separate implementation artifacts from structural limits, test adversarial cases, isolate compression and extraction bottlenecks, and finally turn the method back onto the paper itself in a self-referential recursion test. Across the series, the results support the core conservation claim while also clarifying how observed commitment depends on the interaction between compression, extraction, gating, and evaluation. What emerges is not a collapse of the law, but a clearer account of how conserved commitments appear in practice: sometimes as stable attractors, sometimes as reduced kernels, sometimes as reformulations, and sometimes as apparent failures caused by proxy-layer measurement gaps rather than loss of the underlying commitment itself.

This archive preserves that full record in citable form. It includes narrative logs, tabular reports, machine-readable traces, corpora, and supporting figures for each experiment, allowing readers to follow both the development of the testing program and the reasoning behind each refinement. It is intended to support the main paper without overloading it: the law paper states the principle, while this experimental record provides the empirical lineage behind it. Deeper treatment of harness dynamics, extractor asymmetries, bottlenecks, and edge-case behavior is left to a separate follow-on paper. In this way, the present record serves both as documentation and as an invitation for replication, scrutiny, and extension by others working on commitment, compression, recursive transformation, and semantic stability in language systems.